Intel has always prided itself on its huge capital expenditures. Investments in fabs have helped the company become the world's largest chip manufacturer, as well as the proprietor of the most advanced semiconductor processes.
The chip giant is already making 22 nm chips, when most other chip makers are just starting on 28 nm devices. More importantly, Intel is expected to be the first to move to the 14 nm process later this year.
At the same time, as the PC market continues to struggle, Intel finds itself sitting on lots of extra fab capacity. What to do? The obvious move, but a surprising one for a company that considers the manufacturing process its key competitive advantage: sell it. The company formed a foundry business about three years ago, but until recently its customers have been tiny companies. Its first announced customer, in November 2010, was programmable logic startup Achronix Semiconductor. It signed another startup focusing on programmable chips, Tabula, in February 2012.
This year, however, Intel's foundry effort has finally gained steam. In January, it signed Microsemi Corp., a $1 billion company that sells analog and mixed-signal semiconductors. In February, Intel struck a foundry deal with $1.5 billion Altera Corp.
Brian Krzanich's appointment as Intel's new CEO in May signaled that the company would emphasize and grow the foundry model to take advantage of its strong manufacturing investments and generate revenue, according to analysts. Krzanich came up through Intel's manufacturing ranks. Before becoming COO, he held senior positions in manufacturing, including process technology and also assembly and test.
After decades of touting its manufacturing prowess as its main competitive advantage, it's hard to imagine that Intel can make the foundry model work. It's one thing to make chips for others using older technology, but Intel is apparently opening the kimono for at least one foundry customer. The deal with Altera will give the programmable chip vendor access to Intel's upcoming 14 nm trigate transistor technology, which Altera CEO John Daane believes will give his company a significant edge over its main competitor, Xilinx, according to Reuters.
“We are essentially getting access like an extra division of Intel. As soon as they're making the technology available to their various groups to do design work, we're getting the same,” Daane told Reuters.
The have also been unconfirmed reports that Cisco Systems is an Intel foundry customer.
Intel doesn't have foundry in its DNA, so it will be interesting to see whether it can satisfy its customers and sign on new business. During Intel's second-quarter earnings call on July 17, Krzanich said that Intel would choose its foundry customers carefully. “We'll look at them individually, from a strategic standpoint, looking at what size of business they bring, and whether their products can really take advantage of our process leadership,” he said.
The jury will probably be out for a while. “Even the customers that we've already signed up — the ones you've heard about in the press — those revenues and the impact on Intel's bottom line is still a ways out,” he told analysts.
The company is being careful to put in place the right systems and customer support models that will enable it to operate as a foundry, he said. “Since we haven't been a foundry business, we want to make sure we really know what we are doing.”
What do you think? Can Intel make this shift?
Intel has always prouded itself on producing the most advance chips and selling it to others. If Intel moves into foundry then it will not remain Intel but it will just another TSMC or GF. I think Intel should rather look for a different model where it could find a better usage of its chips and keep producing quality over quantity.
I've always admired Intel for its clear focus–and i understand what's getting them to this place of considering offering foundry services. It will be really interesting to see how it evolves. thanks for the deep dive on this topic.
The problem is that Intel needs the quantity in order to justify its large capital expenditures. It needs to keep those fabs full. And with the PC market waning, unless Intel starts selling a lot more chips into other markets like smartphones and tablets, it won't be able to do that. The alternative is the foundry model. It may not have a choice.
Tam,
What about Automation/Core Manufacturing/Medical Applications?
How is Intel performing there?
I know they got onto the Mobile Game very,very late but does it mean that it really is Game over for them?
After all,Looking at current Trends I see Mobile Sales also falling Globally(same thing happened in 2012 as well as Consumers start to become uber-cautious about Spending their Hard-earned Money as the Recession starts to bite!);also if your Cellphone can work very well for 2 years(average Replacement time for most Western Consumers) maybe it can work well for 3-4 years as well?
After all Cellphones(like any Electronic Item) is a depreciating asset for Consumers;so Why keep wasting you Hard-earned Savings on it?
Regards
Ashish.
Himanshu,
My Sentiments precisely.
Don't you also feel that Medical Industry holds a lot of promise today?
Â
Of course it does. There will always be funding for biotech & research. So it's a revolving door of promising advancements.
“Of course it does. There will always be funding for biotech & research. So it's a revolving door of promising advancements.”
pocharle, i feel lots of scops are there in this type of applications. Technologies are in place, but we have to transform or port such technologies to application levels.
“My Sentiments precisely. Don't you also feel that Medical Industry holds a lot of promise today?”
Tech4people, but i personally feels that medical elctronics sector is one of the promising domain
That process is the difficult part. Turning theory into reality.
“That process is the difficult part. Turning theory into reality.”
pocharle, you are right. Fine turning the devicess in to application for better performance
Performance can be turned into a whole different conversation. It's such a subjective thing that creating “performance baselines” is tough in itself.
“Performance can be turned into a whole different conversation. It's such a subjective thing that creating “performance baselines” is tough in itself.”
Basically software or devices are creating with a minimum performance guarantee. So fine tuning will help to derive maximum out of it.
I know that. But usually you'll find that the 'performance' baseline by most vendors would not be even acceptable in most environments.
The tweaks involved to get the most out of your investment usually requires time and load testing on your own or with the help of experts.
“But usually you'll find that the 'performance' baseline by most vendors would not be even acceptable in most environments.”
Pocharle, that's why companies are giving only minimum performance guarantee and not any guarantee about maximum performance.
I understand that. This is why the intersection between performance & compatibility becomes paramount if you're going by the vendors guidelines.
“This is why the intersection between performance & compatibility becomes paramount if you're going by the vendors guidelines.”
Pocharle, but we have to look for vendor parameters too.
I understand that. I'm talking about the initial stages of the process.
“I understand that. I'm talking about the initial stages of the process.”
Pocharle, thanks for ur understanding.
NP
“NP”
pocharle, NP means?
It's the abbreviation for “No Problem”
“It's the abbreviation for “No Problem””
Pocharle, Thanks for the clarification.
Glad to help.